the problem of blocked jobs you could build Agent Groups from the
agents and specify the agent group as the host for the jobs.
transfer key needs to renewed, the host in
the agent group is not available, so the
workload gets dispatched to the next available agent (according to mode ) in